Mysterious Freeze

Message

Fallen_Paladin · Post by **Fallen_Paladin** » Mon Feb 07, 2005 6:51 am

Heya all,

I was having a similar problem - type 1 freezes - almost immediately upon starting x with kde. After skimming/reading most of this thread, I decided to try the following based primarily on Dave's and Dolio's conclusion that it was glx-related and Cimbom's statement that reverting to version 6111 of the nVidia drivers fixed it for him. Since I didn't have anything emerged dependent upon nvidia-glx-1.0.6629, I unmerged just it (left the nvidia-kernel emerge intact) and moved the files for all versions past 1.0.6111 out of the portage tree. (I moved them from /usr/portage/media-video/nvidia-glx into a temporary folder in my home directory. I doubt this is the right way to manipulate the install, but I'm completely new to this.) I then re-emerged nvidia-glx with the oneshot argument (so that it wouldn't automatically update in the future.) This emerged nvidia-glx-1.0.6111 and currently kde seems to be running fine. After emerging, I replaced the removed files back into the tree. From reading through this thread, it looks like there is more than one problem (based on the variety of fixes), but I'd like to know if anyone else finds this solution to work. (Or, if so, can improve upon this.) Any suggestions about what I should or could have done to achieve a similar result or fix the same problem would also be appreciated.

I'll let you know if it freezes on me again, though, since it seems that some people are waiting quite awhile between freezes. (While I couldn't go more than 3 minutes in kde without a freeze.)

yaneurabeya · Post by **yaneurabeya** » Mon Feb 07, 2005 8:22 am

Fallen_Paladin wrote:Heya all,

I was having a similar problem - type 1 freezes - almost immediately upon starting x with kde. After skimming/reading most of this thread, I decided to try the following based primarily on Dave's and Dolio's conclusion that it was glx-related and Cimbom's statement that reverting to version 6111 of the nVidia drivers fixed it for him. Since I didn't have anything emerged dependent upon nvidia-glx-1.0.6629, I unmerged just it (left the nvidia-kernel emerge intact) and moved the files for all versions past 1.0.6111 out of the portage tree. (I moved them from /usr/portage/media-video/nvidia-glx into a temporary folder in my home directory. I doubt this is the right way to manipulate the install, but I'm completely new to this.) I then re-emerged nvidia-glx with the oneshot argument (so that it wouldn't automatically update in the future.) This emerged nvidia-glx-1.0.6111 and currently kde seems to be running fine. After emerging, I replaced the removed files back into the tree. From reading through this thread, it looks like there is more than one problem (based on the variety of fixes), but I'd like to know if anyone else finds this solution to work. (Or, if so, can improve upon this.) Any suggestions about what I should or could have done to achieve a similar result or fix the same problem would also be appreciated.

I'll let you know if it freezes on me again, though, since it seems that some people are waiting quite awhile between freezes. (While I couldn't go more than 3 minutes in kde without a freeze.)

Well, at least some people have found a rock solid solution to this issue

. I would make sure to mask the nvidia-kernel file so it doesn't update to 6629 because the glx and kernel drivers must be the same. Also emerging the *glx package with the --oneshot flag allows the driver to be up for deletion when depclean is run, just to let you know... So maybe masking that package as well would be a good idea.

Fallen_Paladin · Post by **Fallen_Paladin** » Mon Feb 07, 2005 8:56 am

Actually the nvidia-kernel file is 6629. And the nvidia-glx is 6111. *shrug* This is what worked for me. I'll have to go look up on how to mask something . I've only had gentoo running for a week.

There's definitely more than one problem in this thread (IMO), but it appears much of it revolves around the glx. Any of you that still have it wouldn't happen to have a small partition you could throw gentoo on fresh-like and see if you have the same problem with just nvidia-glx-6111, xorg, and kde would you? I'm curious if other packages (especially xorg glx-using ones) might be exacerbating the problem. Is it possible that a bug or even a bug-fix in the glx is conflicting with some of the (old?) packages that use it?

BTW: I did not have the problem in xorg before I emerged kde. (Mind you, twm isn't much of a test environment, but it worked.)

massimo · Post by **massimo** » Mon Feb 07, 2005 9:03 am

Well, I got these freezes too:

Code: Select all

Feb  7 08:37:41 [kernel] NVRM: Xid: 3, C 00000000 SC 00000007 M 00000184 Data 02050200
Feb  7 08:37:51 [kernel] NVRM: Xid: 3, C 00000001 SC 00000005 M 0000019c Data beff000b

kernel 2.6.10-nitro4
nvidia-kernel-1.0.6629-r3

I want to add some information I didn't found in this thread (to be honest I didn't _every_ post, but most of them in this thread):
I only get these freezes with my MSI Geforce4 Ti 4400 VIVO. When I use my Gainward Geforce4 Ti 4200 I don't get these freezes.

ikaro · Post by **ikaro** » Mon Feb 07, 2005 3:51 pm

tried not using glx, still hardlocks

my xorg.conf is to be found here:
http://fluxmod.dk/files/misc/xorg.conf.html

xorg.log
http://fluxmod.dk/files/misc/xorg.log

dmesg:
http://fluxmod.dk/files/misc/dmsg

emerge info

Code: Select all

Portage 2.0.51-r15 (default-linux/amd64/2004.3, gcc-3.4.3, glibc-2.3.4.20041102-r0, 2.6.11-rc3-bk3 x86_64)
=================================================================
System uname: 2.6.11-rc3-bk3 x86_64 AMD Athlon(tm) 64 Processor 3500+
Gentoo Base System version 1.6.9
Python:              dev-lang/python-2.3.4 [2.3.4 (#1, Jan 25 2005, 18:17:54)]
dev-lang/python:     2.3.4
sys-devel/autoconf:  2.59-r6, 2.13
sys-devel/automake:  1.7.9-r1, 1.8.5-r3, 1.5, 1.4_p6, 1.6.3, 1.9.4
sys-devel/binutils:  2.15.92.0.2-r2
sys-devel/libtool:   1.5.10-r4
virtual/os-headers:  2.6.8.1-r3
ACCEPT_KEYWORDS="amd64 ~amd64"
AUTOCLEAN="yes"
CFLAGS="-march=k8 -O2 -pipe"
CHOST="x86_64-pc-linux-gnu"
CONFIG_PROTECT="/etc /usr/kde/2/share/config /usr/kde/3/share/config /usr/lib/X11/xkb /usr/share/config /var/bind /var/qmail/control"
CONFIG_PROTECT_MASK="/etc/gconf /etc/terminfo /etc/env.d"
CXXFLAGS="-march=k8 -O2 -pipe"
DISTDIR="/usr/portage/distfiles"
FEATURES="autoaddcvs autoclean autoconfig ccache distlocks noinfo notitles sandbox"
GENTOO_MIRRORS="http://mirror.uni-c.dk/gentoo/ ftp://ftp.dkuug.dk/pub/gentoo/ ftp://rsync.gentoo.diku.dk http://ftp.du.se/pub/os/gentoo"
LANG="en_US"
LDFLAGS="-Wl,-O1"
MAKEOPTS="-j2"
PKGDIR="/usr/portage/packages"
PORTAGE_TMPDIR="/var/tmp"
PORTDIR="/usr/portage"
PORTDIR_OVERLAY="/usr/local/portage"
SYNC="rsync://rsync.gentoo.org/gentoo-portage"
USE="amd64 X acpi alsa cdr f77 gtk2 ipv6 ithreads jpeg lzw ncurses nptl nptlonly oggvorbis opengl perl pic png python ssl truetype truetype-fonts usb userlocales xpm xrandr xv"
Unset:  ASFLAGS, CBUILD, CTARGET, LC_ALL

and im about to go nuts ...

Icer · Post by **Icer** » Mon Feb 07, 2005 6:28 pm

Hello folks.

I got the same problem. X opens fine, but try and run glxgears and the screen freezes. I tried some alt sysrq ... combos but I cant get out of X. How o you get out of X? IIRC someone said he got out.

Anyway. I believe my system is stable if I disable SBA, but then I get piss poor performance. Fastwrite wont work with this card it seems. Motherboard supports it but the card doesnt.

Without SBA I get some 70-100FPS in glxgears. If I enable SBA I get +-2300FPS.

I'm using nvidia driver not agpgart. ... Well I've been monitoring these forums and see lots of ppl have similar problems so it must be either nvidia or xorg and I'm beginning to believe it's xorg here which is borked since some ppl with ATI cards have problems too.

carneboy · Post by **carneboy** » Mon Feb 07, 2005 7:54 pm

Just to add myself to the list, I am also having type 2 lockups

occurs after about a day and a half of uptime, usually during a screen-saver or firefox

latest stable nvidia, xorg

keep me posted

Fallen_Paladin · Post by **Fallen_Paladin** » Tue Feb 08, 2005 12:55 am

A couple updates. Using nvidia-kernel & nvidia-glx 6111, my system is still stable after running a glx screensaver overnight then glxgears for awhile.

I tried switching to 6629 and running glxgear in twm, as a test - immediate lockup. And, of course, kde still locks up when opening, too. Note: This IS using the nvidia-glx-1.0.6629, not the nvidia-glx-1.0.6111.

Icer, I have to reboot to get out when it locks up. No safe way to do it. Not fun, but if I try to get out using key-combos, I eventually get a complete freeze.

I think I should also add, since other people have spoken of these in this thread, that I don't have framebuffer support compiled into the kernel and that my filesystem is reiserfs. My video card is an Asus GF4MX4000 64 MB RAM.

If anyone has any ideas about something I should try (in order to see if it will lockup), I'm watching this thread. Or if you have a question about my system, that you think will help.

[Update] I did get a complaint from after one reboot about my nvidia-kernel's version not matching X's version, so I reverted to 6111 on that. It worked until this last reboot though, so I'm not sure what the cause is. Once this happened though, I had to unmerge nvidia-kernel, reboot, remerge 6111, and reboot again. I'm not used to having to reboot my linux machine for a modprobe. In the past I've been running slackware without X, but modprobing and opengl updating didn't seem to work here. X opened, but then froze. Is there alot of rebooting in gentoo or is this very unusual?

Bingo · Post by **Bingo** » Tue Feb 08, 2005 3:54 am

Hi there!
First of all, I didn't read the whole thread, I currently don't have the time. But I did went through a few pages and this is definatly the same kind of problem I have. At first I thought it was an ACPI problem (that's how I found this thread) because it only happens when my screensaver runs but then again, not every time!

For people having to hard reboot their computer, know that if you have ACPI installed and the button module (or built in the kernel), you can at least cleanly poweroff you computer by pressing your computer's power button. I think you have to have the acpi daemon running too.

I run Gentoo in 64 bits and only emerge stable programs.
A few things make me think it's related to X, glx and compositing as you'll see in this list:

-I have Xorg 6.8.0-r4
-I have a Geforce 6800GT running the 6629 drivers
-I only began to experience these lockups after emerging Xfce 4.2 (which went stable on amd64 only a week ago) which has compositing and all the other eye candy on by default.
-As I said earlier, it only happens when my screensaver runs. I run an opengl screensaver, Flurry. In fact, it doesn't happen when my screensaver RUNS, but when it STARTS. I was doing something else yesterday and saw my screen fade to black, when the screensaver kicks in. But the screen stayed black.
-The keyboard doesn't respond to anything (the three lights don't respond to the different LOCK buttons, nor does any combination of key do anything)

I don't know if it is related but I just tried to emerge the tetex package (which is 85MB) twice in a row and went away to do other things. When I came back a few minutes after (long enough for the screensaver to start, but surely not long enough for the compiling to be done), it had the problem again. But there was no hdd activity, so it had stopped compiling... The system still responded to the power button. I don't know if doing some hard work makes the computer more vulnerable to the problem because it usually only do it once in a while, and not twice within 15 minutes.

For GLX, I have the following line in my xorg.conf (I don't know if it could help):

Code: Select all

Option          "RenderAccel" "true"
Option          "AllowGLXWithComposite" "true"

And for Xorg's extension:

Code: Select all

Section "Extensions"
        Option "Composite" "Enable"
        Option "RENDER" "Enable"
EndSection

Bingo · Post by **Bingo** » Tue Feb 08, 2005 4:31 am

Update:

I just tried reemerging tetex and let my computer alone while it was doing it. Soon enough the screensaver came and the problem arose. This is the second time only that I actually see my computer doing the transition from "working" to "frozen".

I ssh'd my computer and realized that I have no X running. I went to read the Xorg log and there's what I saw at the end of it:

Code: Select all

Fatal server error:
Caught signal 11.  Server aborting

And now while writing this I tried to switch console (Ctrl+Alt+1) and actually, my computer is NOT frozen. That's weird because I'm pretty sure I already tried this multiple times It's like a habit, when X fails, I always try Ctrl+Alt+Backspace and then Ctrl-Alt-[1...6] if it didn't work. Looks like I missed it this time...
Anyway, I'm back to the console and can clearly see that the connection to the X server was lost...

What's weird is that previewing the screensaver works and does not make X fail.

I'm clueless.

gspr · Post by **gspr** » Tue Feb 08, 2005 8:12 am

Just a little general update/recap from my side:
Running a GeForce4 on an nForce2 platform with an AMD AthlonXP 2500.
Kernels tried: 2.6.7, 2.6.8.1, 2.6.9, 2.6.10-rc2, 2.6.10., 2.6.10-gentoo-r7.
X.org: 6.8.0.
nVidia drivers: 6629 AND 6611.
Symptoms: X starts to hog 100% CPU time after a seemlingly random(*) amount of time running OpenGL apps (such as UT2004). This causes any OpenGL app to run fatally skippy. The forementioned NVRM lines appear in my kernel message buffer, and all keyboard/mouse input is frozen until I can kill X from SSH.
I have tried with and without: ACPI, APIC, pre-empt and the composite extension.

(*): Extremes: ~1 minute and ~90 minutes.

(I know I've mentioned my problem before, but I thought I'd recap it for readability for the new kids who are lucky enough to be experiencing this problem).

yaneurabeya · Post by **yaneurabeya** » Tue Feb 08, 2005 9:55 am

Bingo: Does your screensaver have OpenGL extensions by chance? Also, try disabling your APM/ACPI because they might be kicking in after a while and you might think that it's the screensaver but it's not.

Bingo · Post by **Bingo** » Tue Feb 08, 2005 8:47 pm

Yep my screensaver has OpenGL extensions. I now have switched to a non-OpenGL screensaver to see if X still crashes.
I doubt that APM/ACPI is the problem, but I couldn't find the kernel option to disable it for testing purpose. I doubt this is the problem because I've never configured ACPI and it it's current (default) state it only seems usefull for catching the poweroff button activity: My screen never goes to standby mode, even after quite a few hours of inactivity (like all day/night long).

yaneurabeya · Post by **yaneurabeya** » Tue Feb 08, 2005 9:14 pm

Hmmm... btw Bingo, from my experience I have issues with power management and you have to completely remove apm/acpi from the kernel or else it will crash, regardless of whether you have the module modprobed or not

...

Speaking of which, I was wondering if someone with devfs might have a set of permissions that I could check for /dev/apm_bios so that I could test out one final thing before I give up for now >_>...

gabio · Post by **gabio** » Wed Feb 09, 2005 3:04 am

/me changed the line:
driver "nvidia"
to:
driver "nv"

in my xorg. and magic, no more freeze.
guess why...

yaneurabeya · Post by **yaneurabeya** » Wed Feb 09, 2005 4:07 am

Could you list your setup (emerge --info, xorg info, nvidia info, and hardware specs)?

Judging by the size of this thread, I think that we should almost make a summary thread to guide people to if they have issues and sticky the jerk.

bunder · Post by **bunder** » Wed Feb 09, 2005 4:16 am

yaneurabeya wrote:Judging by the size of this thread, I think that we should almost make a summary thread to guide people to if they have issues and sticky the jerk.

i made a summary post a few pages ago... please feel free to pm me for corrections or additions.

Bingo · Post by **Bingo** » Wed Feb 09, 2005 5:39 am

Here's my info:

Code: Select all

# uname -a
Linux benoit 2.6.9-gentoo-r14 #1 Wed Jan 12 00:23:53 EST 2005 x86_64 AMD Athlon(tm) 64 Processor 3500+ AuthenticAMD GNU/Linux

Hardware:
Athlon64 3500+
A8V-Deluxe (VIA K8T800Pro)
2x512MB Corsair TwinX
BFG 6800GT OC 256MB

Xorg:
6.8.0-r4

Nvidia:
emul-linux-x86-nvidia 1.0.6629
nvidia-glx 1.0.6629-r1
nvidia-kernel 1.0.6629-r1
nvidia-settings 1.0.6629

And here are the most meaningful parts of my xorg.conf:

Code: Select all

Section "Module"
        Load    "glx"
        Load    "record"
        Load    "extmod"
        Load    "dbe"
        Load    "xtrap"
        Load    "type1"
        Load    "speedo"
        Load    "freetype"
EndSection

Section "Extensions"
        Option "Composite" "Enable"
        Option "RENDER" "Enable"
EndSection

Section "Device"
        Identifier      "nVidia Geforce 6800GT"
        Driver          "nvidia"
#       Driver          "nv"
        Option          "RenderAccel" "true"
        Option          "AllowGLXWithComposite" "true"
EndSection

If you want more of my xorg.conf, just ask, but it's basic stuff about monitor specs, screen resolutions, input devices, files, server layout and that's it. Nothing else fancy.

Also, I don't have any special options enabled for the nvidia driver. Everything is set to defaults.

Thanks for your help!

sammalpa · Post by **sammalpa** » Wed Feb 09, 2005 7:24 pm

Am I correct if I make an assumtion that the actual cause of this freeze is not absolutely clear still ?

I get freeze a full freeze, no mouse, no keyboard, no ssh.. nothing responds.. (type2...)

I have ATI card but only xorg (not ati drivers), 2.6.10_r10 kernel.. usually I thing I have seen it when trying to operate gaim or firefox with mouse..

I have ran debian on this same machine for over a year, and uptimes were constantly weeks.. just lately this appeared on my gentoo box.. Installed it ~2 weeks ago, saw the first freeze 3 days ago..

_d

davecs · Post by **davecs** » Wed Feb 09, 2005 10:45 pm

I've tried downgrading my nvidia drivers to 6111. I'm not holding my breath, though!

yaneurabeya · Post by **yaneurabeya** » Thu Feb 10, 2005 7:28 am

After dealing with some questions, I think that there are separate issues for both us 32-bit users and the 64-bit users. I'm not sure but I think bingo helped provide insight into the fact that the nvidia drivers are off a bit on the 64-bit platform and that by doing that the 64-bit users can once again enjoy their Xorg without crashing.

That still leaves us 32-bit users without a solution though

.

gspr · Post by **gspr** » Thu Feb 10, 2005 6:46 pm

So it's announced that Xorg 6.8.2 is out. Should be interesting to see how it behaves in relation to this once it's in Portage.

ikaro · Post by **ikaro** » Thu Feb 10, 2005 7:47 pm

yaneurabeya wrote:After dealing with some questions, I think that there are separate issues for both us 32-bit users and the 64-bit users. I'm not sure but I think bingo helped provide insight into the fact that the nvidia drivers are off a bit on the 64-bit platform and that by doing that the 64-bit users can once again enjoy their Xorg without crashing.

That still leaves us 32-bit users without a solution though .

64bit here, no nvidia drivers == hardlocks.

voeck · Post by **voeck** » Thu Feb 10, 2005 8:17 pm

Here is another one with type 1 freezes. My system freeze nearly every time I try to emerge anything. But I think I don't have anything OpenGL related running. I have deactivated any screensavers. It also freezes when I kill xscreensaver completely.
Some more details about my machine:
- Pentium 4
- Nvidia Geforce 4 Ti4280 (I try nvidia driver 1.0.6111 and 1.0.6629; no difference)
- gentoo-dev-sources 2.6.9-r9
- xorg-x11-6.8.0-r4

yaneurabeya · Post by **yaneurabeya** » Thu Feb 10, 2005 8:22 pm

gspr wrote:So it's announced that Xorg 6.8.2 is out. Should be interesting to see how it behaves in relation to this once it's in Portage.

We shall see... I've been running 6.8.1 unstable for quite a while now...

Mysterious Freeze

recap..